Submitted via IRC for Bytram
In the introduction to her new book, Hannah Fry points out something interesting about the phrase "Hello World." It's never been quite clear, she says, whether the phrase—which is frequently the entire output of a student's first computer program—is supposed to be attributed to the program, awakening for the first time, or to the programmer, announcing their triumphant first creation.
Perhaps for this reason, "Hello World" calls to mind a dialogue between human and machine, one which has never been more relevant than it is today. Her book, called Hello World, published in September, walks us through a rapidly computerizing world. Fry is both optimistic and excited—along with her Ph.D. students at the University of College, London, she has worked on many algorithms herself—and cautious. In conversation and in her book, she issues a call to arms: We need to make algorithms transparent, regulated, and forgiving of the flawed creatures that converse with them.
I reached her by telephone while she was on a book tour in New York City.
(Score: 0) by Anonymous Coward on Monday November 19 2018, @08:33AM (9 children)
What is this "fda" going to do?
(Score: 3, Interesting) by c0lo on Monday November 19 2018, @10:08AM (6 children)
TFS:
The other way around, i.e. to protect the algos from the flawed creatures that converse with them, then we'd need an EPA for algorithms.
https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by driverless on Monday November 19 2018, @10:26AM (2 children)
What about the pony? Shouldn't there be a pony in there as well? A pink one, with big sad eyes, and possibly wings. And glitter on its hooves.
(Score: 1) by khallow on Monday November 19 2018, @02:16PM (1 child)
(Score: 0) by Anonymous Coward on Monday November 19 2018, @05:23PM
I can only throw in my support if we also have a plank for good dental hygiene, including a law requiring people to brush their teeth.
(Score: 1, Funny) by Anonymous Coward on Monday November 19 2018, @10:33AM (1 child)
Yea, I read that but its nonsense. What exactly are they going to do? Also I skimmed that story and she says her favorite algo is "geoprofiling", so I dont think she knows much about algorithms.
(Score: 3, Interesting) by DeathMonkey on Monday November 19 2018, @06:42PM
What exactly are they going to do?
More specifically, what are they going to do that NIST doesn't do already?
(Score: 2) by suburbanitemediocrity on Tuesday November 20 2018, @02:30AM
Department of Algorithms
(Score: 5, Insightful) by VLM on Monday November 19 2018, @12:24PM (1 child)
Prevent and slow advancement in programming, enhance the profitability of algorithm patents and patent trolling by making alternative algos illegal, increase hardware sales by making certain more efficient algos illegal, increase hiring of middleman and lawyers for regulatory reasons, make free software unaffordable by charging for certification and indemnity, the usual corruption.
(Score: 2) by DannyB on Monday November 19 2018, @05:21PM
Open Source is the reason why this won't happen.
All the big businesses now use open source. So much so that Microsoft Loves Linux now, haven't you heard? We're talking Google, IBM, Oracle, Red Hat, SAP, Amazon, Netflix, and many others, including even Microsoft.
An attempt to make any of that open source illegal or have to pay a troll would draw response from the many other powerful interests who now use open source.
Open Source makes the world go around.
The lower I set my standards the more accomplishments I have.
(Score: 2) by bob_super on Monday November 19 2018, @08:33AM
So many reasons it ain't gonna happen, why even try to troll about it ?
(Score: 1, Disagree) by Anonymous Coward on Monday November 19 2018, @09:18AM (6 children)
It recently came to my attentions that algorithms don't matter. That 90% of programmers forget all the math and formal combinatorics analysis they've learned as soon as they get their diplomas and that the only people buying algorithms books are the academics that never wrote a commercial piece of code in their lives and the math nerds that are kept around at a 1-to-20 ratio of real programmers to "optimize". After much contemplation and deliberation, I concluded that the problem isn't that academia was being unbelievably lazy and irresponsible by never bothering to design safe languages and algorithm templates built-in, but rather, that the government should step in and make the practices of business programming illegal.
Yours, random Phd. algorithms book author in search for tenure/a job at the new agency since I hate writing code and would much rather wank off to math.
(Score: 3, Interesting) by DannyB on Monday November 19 2018, @05:16PM (4 children)
Having written commercial software almost four decades, I find that even for boring business and accounting software the things I learned in CS did and still do matter to this very day. It is important to understand algorithms if you want efficient and correct code.
Examples: I shouldn't use this algorithm because it is just wrong for this application or will have a huge performance impact. I should use that algorithm for that other application because it would be more efficient as well as correct. Always consider correctness before efficiency.
Trivial optimization can happen as you code. (But should not be affecting big design decisions to avoid premature optimization.) An example of trivial optimization is should I use an ArrayList or a LinkedList. Both of these implement the List interface. But they have different performance behavior for different operations. The choice doesn't affect system design, but can impact behavior in a meaningful way. If you're having a performance issue, you profile and optimize.
I would say that Java (and probably C#) programmers would both say we have safe languages with both data structure templates, as well as algorithm templates built in. Because these implement standard interfaces, high quality third party implementations of algorithms and data structures are also available. (Example: Google Guava, and others)
The government should definitely stay out of programming. (Since I usually only speak sarcasm, I recognize yours instantly.) The government is not needed in programming. Leave it to private enterprise to screw things up without government help. They are already much better at it than the government is.
The lower I set my standards the more accomplishments I have.
(Score: 3, Interesting) by DutchUncle on Monday November 19 2018, @05:59PM (3 children)
>>>> Examples: I shouldn't use this algorithm because it is just wrong for this application or will have a huge performance impact.
Having written embedded systems firmware for four decades, I find that most embedded work is done by EEs, and I was the only CS person they had ever met, and they had never learned any of the (what I thought were) basic techniques, AND DIDN'T CARE. Software was the "easy" part. Of course, nobody (least of all me) would have assumed that I as a CS grad should be designing circuitry for 480-volt power systems, but the EEs who were doing that design were assumed to be able to "throw together" the control program for the microprocessor controlling it, and incidentally doing communications on three lines in three different protocols. And while they might have gotten the timing charts right for the hardware, they didn't recognize when coding approaches left them vulnerable to race conditions in the software.
So a finding that "nobody cares about algorithms" would be accurate; a lot of the people doing programming never studied the background of the field and throw code together till it appears to do what they want, on the limited test set of data that they try. (We can discuss testing for error recognition at length another time.)
(Score: 2) by DannyB on Monday November 19 2018, @07:17PM
That's a different world than I live in, but I hear what you're saying.
The lower I set my standards the more accomplishments I have.
(Score: 2) by suburbanitemediocrity on Tuesday November 20 2018, @02:49AM
I used to write code for high power supplies and things rarely got more complicated than
for( ; ; ) {
setpoint = read_voltage( );
DAC_out = k * ( ADC_vin - setpoint ); // the important part
update_display( );
wait_timer( );
}
And yes I've had to do linked lists, searches and a whole host of other CS data structures, but these aren't difficult to pick up in a few minutes of reading example code. Not for power supplies, but larger plant and process control.
(Score: 2) by legont on Tuesday November 20 2018, @03:27AM
I wonder what happens if the device blows and kills a few people in the process. In older world, say civil engineering, the engineer would certify a bridge and go to prison if it falls down with victims. Do we have "bridge designs" with significant software and who takes the responsibility?
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 0) by Anonymous Coward on Monday November 19 2018, @05:26PM
I'm certain the angle of "this will get us more programmers who menstruate regularly" is lurking in here somewhere.
(Score: 5, Funny) by SemperOSS on Monday November 19 2018, @11:18AM (4 children)
The obvious name for this new administration would be ADA, Algorithm and Data Administration!
I don't need a signature to draw attention to myself.
Maybe I should add a sarcasm warning now and again?
(Score: 1, Informative) by Anonymous Coward on Monday November 19 2018, @06:17PM (1 child)
For those who miss the joke, ADA is a programming language developed by DOD specifically to address software quality by building it into the language.
https://en.m.wikipedia.org/wiki/Ada_(programming_language) [wikipedia.org]
(Score: 2) by suburbanitemediocrity on Tuesday November 20 2018, @02:53AM
Ada. Everybody's favorite programming languages.
As a long time Ada programmer, I really liked it.
(Score: 2) by DannyB on Monday November 19 2018, @07:19PM
ADA:
* American Dental Association
* Americans with Disabilities Act
The lower I set my standards the more accomplishments I have.
(Score: 2) by Bot on Monday November 19 2018, @10:12PM
Software
Correctness
Active
Mentoring
is doublespeaky enough for me.
Account abandoned.
(Score: 3, Interesting) by splodus on Monday November 19 2018, @01:36PM (7 children)
I’m sure people are gonna hate this, cos it’s REGULATION!
However, I think she’s got it dead right with her drugs and snake-oil analogy…
Now of course you can take the line that ‘people should be able to mix St Johns Wort with glucose and claim it cures cancer – and the free market will put them out of business!’ But we know that doesn’t happen in reality – people still buy homoepathic ‘remedies’ even when they know for a fact they are just buying lactose pills and nothing else. Because ‘water has a memory...’ etc – some people will believe anything if the Government does not regulate against bogus claims.
It’s going to slow down innovation, some innovations won’t happen cos the process will be too onerous. It risks legitimising software patents or some similar new intellectual property legislation. There will be a can of worms over the definition of ‘algorithm’ at least.
I’m sure most of us here remember the legal challenge to software in speed-gun devices? How do we know the speed-gun is correct if we can’t see the code?
She uses the example of court-room AI to decide on bail cases. It’s only a matter of time before companies are selling expert systems that claim to make sentencing ‘objective’ - and ultimately assist juries in deciding guilt. Do we want those systems to be secret? Free from regulation?
It’s easy to hand-wave away concerns about Facebook influencing public opinion – but if a totally unregulated, opaque ‘algorithm’ is deciding whether or not you (or your son) is guilty of rape? Who will be comfortable with that?
And I think that is a good example – because there is tremendous pressure right now to ‘increase the number of guilty verdicts in rape cases’. A tweak of the threshold parameters in the algorithm will deliver that – but who will oversee the tweaking?
It’s going to be fraught, of course – but it’s people like us, who have an inkling of the problems AI and Expert Systems might suffer? Surely we’re the people who need to consider this issue? ‘The Public’ have trust in computers – that they are entirely objective and immune to bias!
I can’t help thinking that if we don’t get this regulated right now, it’s going to be decades before the worst consequences of flawed systems gain any recognition.
There’s a hell of a lot of damage could be done in the meantime, even if the free market sorts it all out in the end...
(Score: 3, Insightful) by khallow on Monday November 19 2018, @02:30PM (2 children)
What makes you think such programs are free from regulation now? What exactly would be made non-secret by a FDA-style organization that would be useful?
I hate this not because it's regulation, but because it's particularly stupid regulation. It'll be great for slowing things down, derailing productive economies, creating secure oligopolies (like Big Pharm), and just being yet another hoop of meaningless idiocy that businesses would have to jump through in the course of trying to function. There's no problem to address, and we already have a strong indication that the real world FDA works awfully with billions of dollars gambled away on various chemicals and medical procedures, often eventually rationalized only by some dubious p-hacking.
aristarchus complained in his recent journal [soylentnews.org] about "sociopaths" opposed to ethics. This is really how you create such opposition. Here, we have an ethics proposal that is great for controlling various parties, but useless for fixing actual ethical problems, while committing considerable harm in the process.
(Score: 2) by splodus on Monday November 19 2018, @05:17PM (1 child)
Thanks for that.
You are quite right – I’ve assumed there’s no regulation just cos I’m not aware of it, and she hasn’t mentioned any, but works in the field – so I’m just guessing she would have said if so... If there is regulation that you’re aware of, I’d be grateful if you’d link to some info?
I don’t think we can pronounce her thesis as ‘stupid’ till we know more – not sure I can be arsed to read her book, but it is something that interests me, so maybe I’ll find the time! (I’m hoping there’s more detail than just ‘we need an FDA!’ - but also, highlighting a problem doesn’t necessarily require a solution to make it legitimate, surely?)
I absolutely agree with you that regulation is going to cause its own problems – and I also agree that big pharma has benefited from the FDA’s systems at the expense of small operators and innovation in general…
Would you favour leaving things as they stand then? Leave it to the Free Market?
My default position is to avoid regulation whenever possible, so I’ve got a lot of sympathy with your view. I certainly wouldn’t want to see a situation where no one could write software without jumping through a load of legislative hoops!
There’s a danger of ‘something must be done!!!’ It’s a recipe for unintended consequences. I can’t say existing laws regulating medicines have definitely had a net benefit; maybe we’d have a cure for cancer now if it wasn’t for the FDA?
At the very least this must be a debate worth having?
I guess my fear is that the vested interests will object, of course – and they are powerful indeed. If the tech community doesn’t speak out about the dangers of relying on AI, then who will?
The media is making money off this tech – the big money is consolidating its position with it – so if the Courts become dependent on it, is there any avenue left to challenge the outcomes, in principle?
(And sure, you could argue that ‘people won’t stand for flawed judgements! They will vote out those who support it! Democracy!' But here in the UK – we are about to jump off a cliff as a direct result of the smart use of opaque algorithms to sway public opinion...)
(Score: 1) by khallow on Tuesday November 20 2018, @01:39PM
For example, there's plenty of cases where technologies or particular practitioners of technology have been ruled out as court evidence. So for the algorithm that's supposed to be used for sentencing? Show bias of the illegal sort (such as against gender, religious beliefs, etc in the US), and your have the basis for overturning every bit of sentencing done with that algorithm.
Scratch the surface of almost any human activity and someone is regulating it. You just need to look.
It still needs to be a problem in the first place. The key flaw is simply that an algorithm is not an action. When you regulate algorithms, you aren't actually regulating the problem behavior. And as I noted earlier, there's already regulation (and means to implement more regulation should that become necessary) that don't require any sort of specialized regulatory system. People have been implementing bureaucratic algorithms, for example, for thousands of years.
Further, most of the problems mentioned would be problems no matter what the algorithm was. Such as the project that siphoned a couple million UK residents' health data.
What AI? It's not what we have now. And we don't know enough, in the absence of credible AI, to decide what aspects need regulation.
(Score: 0) by Anonymous Coward on Monday November 19 2018, @05:03PM (3 children)
you don't have to set up a parasitic, unconstitutional bureaucracy, ffs. just make a law that says all government purchased software must be Free Software. done.
(Score: 2) by splodus on Monday November 19 2018, @05:51PM (1 child)
I wouldn’t object to that – but is it realistic? Would a law like that ever get passed in the US? I don’t think it could happen in the UK…?
(Score: 0) by Anonymous Coward on Tuesday November 20 2018, @04:09AM
Not until the working class seizes political power and gains democratic control over the major industries, including software. Proprietary software would fade away after we open source all of the major software companies' code.
One would also hope such a system would prevent the kind of software developer only possible through private sponsorship (*coughpoetteringcough* pardon me, just some symptoms) from ever entering the field or at least prevent him from gaining any sort of influence over anything important.
(Score: 2) by deimtee on Tuesday November 20 2018, @02:06AM
You would never get that, but you might be able to argue for all software purchased by the government must include a full copy of the source code and build environment. You are not prohibiting anything, just ensuring that it is auditable and maintainable.
You would need to come up with a title that acronyms to SECURE or SAFE or PUPPIES or something.
PUPPIES would be good, you could ask opponents "Why do you hate puppies, are you some sort of monster?".
If you cough while drinking cheap red wine it really cleans out your sinuses.
(Score: 2) by Bot on Monday November 19 2018, @01:57PM
no peon, you should only run code vetted by a trusted authority that can do you no harm.
Account abandoned.
(Score: -1, Troll) by Anonymous Coward on Monday November 19 2018, @02:19PM
Only a sick jewish mind could come up with more regulation and central control. Remember how they love communism and hijack all good things and corrupt them.
A person writing an algorithm (walking their dog | using a pencil | repairing their electronics | repairing their home | ...) answerable to authority is their goal. And the only way they could do any work would be if they get approval from the jewish high commissioner for algorithms and data sciences. I'm sure they already have a post for such a thing.
They can organize and plot, but their wishes and hopes will go up in flames, just like them.
(Score: 1, Interesting) by Anonymous Coward on Monday November 19 2018, @03:20PM
is pretty arrogant. I can say so, from having done too much of it over the years.
The FDA is an apologist for the health insurance industry, not a regulatory agency. Same goes for the EPA, DOE, and FCC. Only once in a while do they actually come up with standards that create interoperability between manufacturers or service providers. Don't take my word on it, look at their actual budgets and how they operate.
The question isn't how to administrate the code, it is how to encode the administration. Which is mostly a matter of changing the view of "governance" from one where administrators are expected to come up with ideas, to one where administrators are expected to select from some sort of ISO style standards that are created by the scientific community. Because you can't administrate something that you don't understand at some fundamental level. Before the legal or programmatic solutions can be spec'd out, you have to know what platform you're running on.
The law isn't required to compile. Regardless of how the law is written, the way it operates diverts quite a lot by the time it reaches the bench. IOW, it doesn't have to make empirical sense as a whole, and so it is self conflicting in many places. The judiciary just makes things up to fill in the gaps, in an attempt to give the appearance of homogeny in a fundamentally uncohesive, badly errored, poorly designed system. The regulators (FDA etc.) generally works against clearly interpretable legislation in order ot protect special interests.
The lesson that should be learned from that, is that the only way to regulate is to litigate. Which is to say that if an algo is doing something bad, the purveyors of that algo need to be brought to justice. That is the only reliable mechanism for creating lasting regulatory oversight in this country. You're next question should be: "under what law?". And that is quite simple, because if you have experienced a loss as the result of malfeasance, you can bring a civil suite. It is the judiciaries responsibility to interpret the law. If there is no specific law, then the more general ones apply. You loss still demands reperations, and the courts may assign them to you.
If you look at the law from a programmers viewpoint; the statute isn't the code. The bench law is the code. The statute is badly written API documentation. Which is to say the regulations come after the lawsuits, not before.
(Score: 3, Funny) by darkfeline on Monday November 19 2018, @08:45PM (1 child)
Algorithms are math, what's there to regulate? You cannot make bubble sort run in O(log n) worst case any more than Indiana can define the value of pi through legal fiat.
I mean, I guess you could make using bubble sort illegal, punishable by death, but where's the fun in that?
What I see is someone who doesn't know the difference between computer science and computer engineering and hence her opinions are of little value.
Join the SDF Public Access UNIX System today!
(Score: 2) by legont on Tuesday November 20 2018, @03:31AM
Don't they try to regulate crypto - a tiny math subject?
"Wealth is the relentless enemy of understanding" - John Kenneth Galbraith.
(Score: 2) by corey on Monday November 19 2018, @09:32PM
As above
(Score: 0) by Anonymous Coward on Monday November 19 2018, @11:34PM
The world needs less millennial nitwits like you. If you spent more time teaching your students C, and less time teaching them how to code gender-selection algorithms in Ruby on Rails, the world would be a better place.
(Score: 2) by hendrikboom on Tuesday November 20 2018, @01:03AM
The big question is that of bias. Are the algorithms that determine what you get to see biased in some significant and not widely known manner? I mean, if I connect to a search site about category theory, and it shows me social media posts about category theory, that might be OK. But if I connect to a general news site and ask about the Aswan dam and it shows me nothing but category theory, I would suspect I'm getting biased news.